How to connect to Spark on Mesos outside of HDP from Tableau?

[#1]

A customer would like to explore using Tableau connected to SparkSQL. Their Spark cluster runs on Mesos and their data for this experiment are just plain json files stored on AWS S3. They don't want to move the files in HDFS and they don't want to leverage any service from the HDP cluster to run the ThriftServer needed for the Tableau ODBC connection to Spark. These files will be mined for some value and discarded after a week.

[#2 · Accepted solution]

Hoping you have completed all the pre-requisites to run spark on Mesos, however please follow below if you haven't done yet.

http://spark.apache.org/docs/latest/running-on-mesos.html#connecting-spark-to-mesos

Regarding spark + Mesos and Tableau connection, I believe you need a SparkSql thrift server so that Tableau can directly connect to the thrift port. Morever you can start your thrift server like below.

$SPARK_HOME/sbin/start-thriftserver.sh --master mesos://host:port --deploy-mode cluster --executor-memory 5G

Note: You also need spark ODBC driver at Tableau client side to connect to the Thrift server, you can download it from Here